A Study of the Images Classification on the CIFAR10 Dataset Based on CNNs

Xuanyi Shen

2023

Abstract

Realizing the purpose of identifying and classifying the images relies on machine learning and deep learning methods. It is about building the model and training and testing it. The model calculates the data in the layers and nerve cells inside and creates the most suitable links and relationships between the images and labels. According to the results summary and evaluation indicators, the parameters are adjusted to get the most ideal optimization algorithm, with the highest accuracy and efficiency and least loss. In this paper, a Convolutional Neural Network (CNN) model is built and used to work on the Cifar10 dataset. Its mission is to successfully divide the pictures in the testing set into 10 classes, after being trained by the pictures in the training set and find the most workable algorithm after adjusting to see how well CNNs indeed do while operating the visualizing materials. About the results, it is easy to tell that this method is of great success. The accuracy of it reaches as high as 87.29% while testing, with only the loss of 0.39. Additionally, the efficiency of it is also high enough. To make the conclusion more scientific, this model is compared by the Naïve Bayes model, and the CNN performs apparently rather better than the traditional ones when facing such complex data. Thus, there is the conclusion that the CNN methods are quite capable of work of identifying and classifying the images.

Paper Citation

in Harvard Style

Shen X. (2023). A Study of the Images Classification on the CIFAR10 Dataset Based on CNNs. In Proceedings of the 1st International Conference on Data Analysis and Machine Learning - Volume 1: DAML; ISBN 978-989-758-705-4, SciTePress, pages 114-118. DOI: 10.5220/0012798800003885

in Bibtex Style

@conference{daml23,
author={Xuanyi Shen},
title={A Study of the Images Classification on the CIFAR10 Dataset Based on CNNs},
booktitle={Proceedings of the 1st International Conference on Data Analysis and Machine Learning - Volume 1: DAML},
year={2023},
pages={114-118},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012798800003885},
isbn={978-989-758-705-4},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Data Analysis and Machine Learning - Volume 1: DAML
TI - A Study of the Images Classification on the CIFAR10 Dataset Based on CNNs
SN - 978-989-758-705-4
AU - Shen X.
PY - 2023
SP - 114
EP - 118
DO - 10.5220/0012798800003885
PB - SciTePress